diff --git a/.vscode/settings.json b/.vscode/settings.json index 6384aad..202338e 100644 --- a/.vscode/settings.json +++ b/.vscode/settings.json @@ -1,5 +1,6 @@ { "cSpell.words": [ + "civilizational", "DALL", "HITL", "Nightcafe", diff --git a/content/posts/2024-god-send-me-a-sign/index.md b/content/posts/2024-god-send-me-a-sign/index.md index 20321d6..9743794 100644 --- a/content/posts/2024-god-send-me-a-sign/index.md +++ b/content/posts/2024-god-send-me-a-sign/index.md @@ -1,6 +1,6 @@ --- title: I need to see the Sign - destroy that truck! -description: We must realize that everyone experience is unique on this earth, and must respect and accomodate them as much as possible +description: We must realize that everyone experience is unique on this earth, and must respect and accommodate them as much as possible author: Adi Rabinovich type: post heroStyle: "background" diff --git a/content/posts/2024-search-transformation-in-AI-age/featured_fantasy-holographic-pop-up-book.webp b/content/posts/2024-search-transformation-in-AI-age/featured_fantasy-holographic-pop-up-book.webp new file mode 100644 index 0000000..a043b97 Binary files /dev/null and b/content/posts/2024-search-transformation-in-AI-age/featured_fantasy-holographic-pop-up-book.webp differ diff --git a/content/posts/2024-search-transformation-in-AI-age/gzILk0VPG5fS9811KkTP--1--rqydr.webp b/content/posts/2024-search-transformation-in-AI-age/gzILk0VPG5fS9811KkTP--1--rqydr.webp new file mode 100644 index 0000000..5f87d87 Binary files /dev/null and b/content/posts/2024-search-transformation-in-AI-age/gzILk0VPG5fS9811KkTP--1--rqydr.webp differ diff --git a/content/posts/2024-search-transformation-in-AI-age/index.md b/content/posts/2024-search-transformation-in-AI-age/index.md new file mode 100644 index 0000000..84e0b30 --- /dev/null +++ b/content/posts/2024-search-transformation-in-AI-age/index.md @@ -0,0 +1,83 @@ +--- +title: Search transformation into AI Agents - Today (2024) and Tomorrow (2025+) +description: Adi Rabinovich musings on state of the search, including Google, Perplexity, OpenAI, Claude and more +author: Adi Rabinovich +type: post +heroStyle: "background" +showTableOfContents: true +draft: false +date: 2024-12-24T11:11:11+00:00 +url: /2024/search-transformation-in-AI-age +tags: + - Innovation + - Artificial Intelligence + - Google + - Microsoft + - Perplexity + - OpenAI + - Anthropic + - RAG + - RIG + - AI Age + - LLM + +--- +## Overview + +It has been a breakthrough year in Artificial Intelligence, and many industries are already transforming. Interestingly, Seems like Google missed the importance of their own AI research! Still - I would like to argue that No Other Company Today is better positioned to replace Search with AI agents. The closest competitor, in my opinion, is Apple - although currently their AI is really struggling. + +And the reason for that is - **DATA**! While this is not a surprise - I believe that Google's long standing mission to organize world information is why they have such a huge lead. Let's dive in! + +## AI Agents - Is that not what we already have? + +Key difference in evolution from current AI to proper Agents is Context and Executive Function. Today even the "simple" ability to recall relevant context of our "conversations" with Gemini or Claude represent major challenge for the industry. Proper Agent needs to retain All Key Elements of "conversational awareness" separating long-term and short-term context, as well as accounting for exceptions and evolution. + +Even more distant is safe and effective Executive Function - Where we will all be confident enough to let AI Agent actually Buy our flights, or even filter down to top 3 hotel options (explaining key Relevant differences) - to stay with a Travel Agent example. + +## Google Search, AI Agents and the Path Forward + +Although Google made most significant AI investments over the last 15+ years, they managed to fumble them also, as I believe no company is immune to the famous [Market Cannibalism Conundrum](https://www.investopedia.com/terms/c/corporatecannibalism.asp#:~:text=Corporate%20cannibalism%20is%20when%20a,product%2C%20therefore%20reducing%20overall%20sales.) (More on this under a Tangent below). + +It is important to understand that LLMs are modeled after human-brain, and as such - they exhibit similar limitations best described by Kahneman in [Thinking, Fast and Slow](https://en.wikipedia.org/wiki/Thinking,_Fast_and_Slow). In other words - LLM has access to "basic facts" about our world, but no access to detailed, real-time, numerical and a lot of other Data about the NOW. + +This is where [RAG](https://cloud.google.com/use-cases/retrieval-augmented-generation?hl=en) and [RIG](https://blog.google/technology/ai/google-datagemma-ai-llm/) come in, but even more so, advanced AI agent should be able to dynamically activate ANY external API! In other words - combination of Core Intelligence with Access to World's (Real-Time) Information opens up the most exciting possibilities. + +Real World Example: You will be able to ask Google AI to "Schedule an appointment with Best Local Children's Eye doctor that takes our insurance and has earliest available appointment!". AI Agent will be able to understand the natural language request, and then access Local Doctors, their ratings, specialties, insurance acceptance And eventually schedule via real-time Appointments API! What I am arguing is that No Other Company is closer to being able to access/obtain so much information about our "Real-World" except Google. Many others can try to implement something similar, but even so will remain at the mercy of [Google APIs speed/cost](https://www.forbes.com/sites/barrycollins/2023/06/01/death-by-api-reddit-joins-twitter-in-pricing-out-apps/). + +## Tangent: Ride the "Google Wave" - or NOT + +The subtitle is of course referring to the famous [Google Wave](https://en.wikipedia.org/wiki/Google_Wave) innovation, which undoubtedly was ahead of it's time. Many of us recall that Google used to be in constant innovation mode, but over time they realized that some "Mistakes" have much higher cost than others. Whenever products (remember Google+?) or major technology frameworks get retired, the company started to realize that it may be hurting their brand to **Market-Test** New Innovative Ideas. Many people (especially important early-adopters) are susceptible to emotional scarring from discontinued tech, which ultimately accumulates. That probably deserves a whole separate post, but the reason I bring it up here is because I suspect this was the main factor stifling AI progress at Google during the (critical) 10+ recent years. + +My sense is that they got rude awakening in 2023/24 and finally managed to Ease the internal guard-rails blocks, which is another critical Factor that should make Google better positioned in this battle. Their enormous pace of [innovative releases in 2024](https://blog.google/products/gemini/google-gemini-ai-collection-2024/) I believe is actual evidence that they managed to (correctly IMHO) outgrow their **Brand-Trauma**. + +## Apple Agent still MIA (Siri team struggles) + +It is not a secret that [Siri has struggled](https://news.ycombinator.com/item?id=16588645) for years. I suspect current sad state of affairs with Apple AI is a result of the same team leading the effort - they really need a refresh (Call me?) there! Now to be more specific - While I Applaud [Apple's pioneering efforts in AI Privacy](https://security.apple.com/blog/private-cloud-compute/), I am not sure the resulting delay in bringing AI to the masses was worth it during such a critical juncture of our civilizational entry into the AI Age. + +## Another Tangent: Technology Positioning + +While Market Positioning for a Brand seem to be a well defined metric, what I am referring to here is more of a Technology positioning indicators. In other words - there are serious Platform Migration hurdles consumers need to overcome, and as they develop habits and expectations, sprinkled with brand loyalty, this barrier grows almost impenetrable over time. +Of course this is the position both Google and Apple enjoy in our lives at the moment - something even mighty [Microsoft failed to penetrate](https://www.slashgear.com/1643513/why-microsoft-discontinued-windows-phone/). Thus I postulate that these two companies will remain at the forefront of Search Evolution to AI Agents, building upon their Mobile Assistant/Platform dominance. + +## How is OpenAI and Perplexity Faring, and others + +This existential Mobile-Platform-Reach risk is also why OpenAI, Perplexity and many others adopted Mobile-First strategy - but is having their own App on our phones sufficient? I fear I lost track several times of the many AI App icons I have on my phone - I eventually created a dedicated AI Agents group for them, further relegating them to obscurity within that "group of tiny icons" paradigm... + +The other huge hurdle of course is Data Access: as most AI companies are discovering, [training data has value](https://www.wsj.com/business/media/wall-street-journal-new-york-post-sue-perplexity-ai-c5d9554d). And of course these litigations are really a precursor to monetizing such information-pipeline. Again, this is where Google excels, as with proper (fun) Gamification they managed to get all of us to [Love helping them collect real-time data](https://medium.com/@krishnan357/google-maps-an-example-of-gamification-done-right-2e5a4a541add) about Our World! Finally I already wrote above about the API challenges - even if it works today, I would consider it a sort of "[free public preview](https://arstechnica.com/information-technology/2011/10/google-warns-that-rate-limits-overage-fees-are-coming-to-maps-api/)"... + +## Conclusion + +Interesting and Exciting Times Ahead!!! We truly live in an amazing age of humanity transformation! "AI Privacy Leader" seems to be how Apple is trying to differentiate itself in this battle - hoping this will tip the scale on ultimate consumer decision of **Which Agent To Use**. Judging from the past - most of us are willing to sacrifice Privacy for further Convenience, however... + +![AI Robots Self Exploration](./gzILk0VPG5fS9811KkTP--1--rqydr.webp) + +As I was writing this - [OpenAI announced groundbreaking scores on ARC-AGI-1 set](https://arcprize.org/blog/oai-o3-pub-breakthrough). If you are following closely - o3 is their "chain-of-thought" reasoning system, designed from the ground up to slowly explore various Steps towards optimal solution, instead of quick matter-of-fact responses only. + +## Art Credits and Additional Reading + +- Featured AI Art by [George Stanley via Nightcafe](https://creator.nightcafe.studio/creation/8EAQvCqBG4truXC5f6Gs/fantasy-holographic-pop-up-book?ru=adir1) (Flux PRO 1.1 Ultra) +- Robots AI Art by [BillThegutterRat](https://creator.nightcafe.studio/creation/gzILk0VPG5fS9811KkTP?ru=adir1) (DSXL 0.9) +- [Google Preparing for Post-Search Era](https://www.linkedin.com/pulse/google-preparing-post-search-era-rafael-goldschmidt-xx49e/) +- [Anthropic - Building Effective Agents](https://www.anthropic.com/research/building-effective-agents) +- Alas - we didn't talk about [Hallucinations](https://cloud.google.com/discover/what-are-ai-hallucinations?hl=en) Here! +- [OpenAI - Learning to Reason with LLMs](https://openai.com/index/learning-to-reason-with-llms/) diff --git a/content/posts/2024-study-of-borders/index.md b/content/posts/2024-study-of-borders/index.md new file mode 100644 index 0000000..7b6bf0f --- /dev/null +++ b/content/posts/2024-study-of-borders/index.md @@ -0,0 +1,76 @@ +--- +title: A Study of Borders +description: Adi Rabinovich musings on impact of borders in our pursuit of World Peace for human civilization +author: Adi Rabinovich +type: post +heroStyle: "background" +showTableOfContents: true +draft: true +date: 2024-10-21T11:11:11+00:00 +url: /2024/study-of-borders +tags: + - Musings + - Research + - Politics + - Civilization + +--- +## Brief + +Borders and Walls appear to be essential element of our society. We see them everywhere - in our homes, and yards and streets. We also encounter them on a bigger scale when crossing into other cities, or county lines, and especially with state/country borders. Yet my sense if few understand their symbolic significance and most importantly - what is the power dynamics around them. Here I attempt to reflect on this important concept, especially with their anthropological significance for our entire civilization - as an obstacle/solution to World Peace. + +## History + +Genghis Khan + +### Gemini + +**The Great Wall of China had a mixed impact on Genghis Khan's attack.** + +**On one hand, the Great Wall served as a formidable obstacle.** It was a massive fortification, stretching thousands of miles and often equipped with watchtowers and garrisoned by soldiers. This made it difficult for the Mongols to bypass or overrun. + +**However, the Great Wall also had its limitations.** It was not a continuous structure, and there were gaps and weaknesses that the Mongols could exploit. Additionally, the Mongols were skilled at siege warfare and had developed tactics to overcome fortified defenses. + +**Ultimately, the Great Wall did not prevent Genghis Khan from conquering China.** While it slowed his progress and caused him losses, it was not insurmountable. The Mongols' superior military tactics, discipline, and determination allowed them to overcome the wall and ultimately establish their dominance in the region. + +### ChatGPT + +The **Great Wall of China** was not a direct reason for Genghis Khan’s invasion of China, but it did symbolize a broader set of circumstances that contributed to his motivation to attack the Chinese states. Several factors linked to the Wall, and the overall geopolitical situation in China, played into his decision-making. Here’s how the Great Wall and its context fit into the larger reasons for the invasion: + +### 1. **Symbol of Wealth and Resources** + The Great Wall represented a long history of wealth, power, and resources of the Chinese empires. Genghis Khan and the Mongols were aware of China’s vast riches, both in terms of land and material wealth. The Wall, though intended as a defensive structure, was a visible symbol of these resources. It may have reinforced Genghis Khan’s ambition to conquer and control the prosperous Chinese territories. + +### 2. **Desire for Tribute and Trade** + Genghis Khan initially sought to establish peaceful trade and tribute relationships with the Chinese dynasties, particularly the **Jin dynasty** that ruled northern China. The Wall was part of the defensive posture of these dynasties, often closing off opportunities for trade or diplomacy. The Wall, as a barrier, reflected China’s historical attempts to manage and control relations with nomadic peoples like the Mongols. When the Jin dynasty rebuffed or limited trade opportunities, it gave Genghis Khan further motivation to use force to achieve his aims. + +### 3. **Perception of Insult and Resistance** + Genghis Khan’s decision to invade was partly motivated by perceived insults and dismissive attitudes from Chinese rulers. The Wall, as a defensive fortification, might have symbolized the **Jin dynasty's** perceived superiority and their desire to separate themselves from the nomadic Mongols. This division may have fed into Genghis Khan's ambition to challenge their authority. When diplomacy failed and tribute demands were unmet, it fueled his desire for conquest. + +### 4. **Strategic and Political Goals** + The Wall also stood between the Mongols and their strategic objectives in China. Genghis Khan aimed to subjugate neighboring powers, expand his empire, and establish control over important trade routes. Since the Wall was designed to protect northern Chinese states like the Jin dynasty, conquering these territories became part of his grander strategy of empire-building. The Wall, in this sense, was an obstacle to his larger goal of uniting and controlling both the nomadic and settled societies of Asia. + +### 5. **Mongol Tradition of Raiding** + Nomadic groups like the Mongols had long engaged in raids against settled agricultural societies, and walls were often built to defend against such incursions. The Great Wall’s presence, while defensive in nature, could be seen as a challenge to the Mongols’ traditional raiding lifestyle. In many ways, it represented the ongoing tension between the nomads of the steppe and the sedentary societies of China. + +### Conclusion: +While the Great Wall of China itself wasn’t the **direct** cause of Genghis Khan’s invasion, it embodied the barriers—both literal and figurative—that stood between the Mongols and the wealth, resources, and power of China. The Wall symbolized the Chinese empires’ desire to protect themselves from nomadic threats, but it also became a target for Mongol expansion. Genghis Khan’s invasion was driven by a combination of economic, political, and cultural factors, with the Wall representing the divide between the Mongol steppe and Chinese civilization. Thus, while not a primary cause, the Great Wall was part of the larger context of Genghis Khan’s motivations. + +### Conclusion-Alt: + +While the Great Wall of China was a significant defensive structure, its presence or absence would not have fundamentally changed Genghis Khan's desire to conquer China. His motivations were deeply rooted in economic opportunity, political fragmentation in China, and his ambition to build a vast Mongol Empire. The Wall was more of an obstacle than a cause for the invasion. Even without the Wall, the factors that drew Genghis Khan to China—wealth, power, and strategic dominance—would have remained, making an invasion almost inevitable. + + +## Anthropology + +I'll admit - when I thought of the title, I wasn't aware there is a whole Degree named that: +https://catalog.csusb.edu/colleges-schools-departments/arts-letters/english/border-studies-minor/ + + +## Conclusion + +Ultimately the wall/border concept is undoubtedly very useful for any society. However of most important to understand is that the border doesn't solve anything per-se. In today's politics there are many people who "want secure borders", and to hell with the whole world burning outside. + +the border itself provides no power - it is legal/military societal structures that give it power. For example - it is great to have a wall between apartments, however disputes around noise, wall disrepair, and more - all persist and often require higher authority/power to resolve. + +Thus it is important to reflect on world stage as a cohesive organism of "dictatorships/democracies" and much more - to make for a safe planet. In other words - unilaterally sitting at home and "securing borders" will ultimately lead to much bigger wars and destruction down the line. + diff --git a/content/posts/2025-inauguration day-thoughts-on-human-rights/index.md b/content/posts/2025-inauguration day-thoughts-on-human-rights/index.md new file mode 100644 index 0000000..b77f0f8 --- /dev/null +++ b/content/posts/2025-inauguration day-thoughts-on-human-rights/index.md @@ -0,0 +1,73 @@ +--- +title: Human Rights, Individual Freedoms - an Antithesis to Dictatorship +description: An exploration of how can people hold Individual Freedoms and Human Rights as paramount while also believing that Dictatorship governance is viable or even efficient option +author: Adi Rabinovich +type: post +heroStyle: "background" +showTableOfContents: true +draft: false +date: 2025-01-20T11:11:11+00:00 +url: /2025/inauguration-day-thoughts-on-human-rights/ +tags: + - Markets + - Economics + - Quants + - AI + - Derivatives + - Options + +--- +## TLDR + +I discovered the amazing world of algorithmic trading probably about 10 years ago. I was already enjoying the exploration of market dynamics, their correlation with human psyche, economic cycles, and companies hype/performance. This post is my musings about AI impact on Black Box algo-trading, and even if one perfects their AI to make endless profits - is it ethical? No prior Quant expertise required to participate, but basic stock market understanding is a plus! + +## Introduction to Algorithmic Trading and Quants + +At its core, the idea of algorithmic trading is pretty straight forward. A computer system is programmed with certain rules and empowered by a human to analyze certain Financial market data and decide which stock to buy or sell, and when. It gets exponentially more complex quickly, though, as most strategies (aka: algorithms/rules) involve [trading Options](https://www.investopedia.com/terms/o/optionscontract.asp) also. + +The Math Wizards use statistical analysis of a lot of Past data, to try and predict future behavior of the market. One of the best known and oldest such trading funds is [Renaissance Technologies LLC](https://www.rentec.com/). Founded in 1982 they boast 90 PhDs in Mathematics, Computer Science and related fields. They also mention that: "Our research database grows by more than 40 terabytes a **day**"! + +So what do they do with all that data? They try to keep their algorithms a secret, but at the core it is a lot of Machine Learning, Reinforced Learning, Decision Trees and even Deep Neural Nets with [Backpropagation](https://en.wikipedia.org/wiki/Backpropagation). Arguably this is actually one of the Easier use-cases - as you have Very specific goal, Maximize Profits, whereas generalized LLMs often operate in murky opinion-based world (especially now in our [post-truth society](https://en.wikipedia.org/wiki/Post-truth)). + +## Dark Pools - What are those? + +So where are the promised Dark-Pools you ask? I would argue that is a renegade semi-legal branch of algo-trading. It is utilizing Level-2 quotes data to look at Live market orders and predict the sentiment for specific stock or market overall. The idea is pretty basic - if there are a lot of sell orders, and fewer buy orders - clearly stock is bound to go lower in the short-term, and vice-versa. This is usually only useful for high-speed trading, also known as day-trading, but clearly whoever has the fastest computer (and market connection), should be able to "predict" the short-term price swing and trade fast enough to take advantage of that. + +This is where our Dark-Pools come in - Large Institutional Funds also have to buy/sell stocks, but obviously they will disproportionally affect the market due to the Large size of their positions (usually **multi-million** dollars) - so it is a Huge signal of *Sentiment* for a stock or even an entire market. In a way [Dark-Pools emerged exactly to Avoid exposure](https://www.investopedia.com/terms/d/dark-pool.asp) of such large participants, but then our human instinct of [Have The Cake and Eat It](https://en.wikipedia.org/wiki/You_can%27t_have_your_cake_and_eat_it) kicked in! + +## Ethics of *Market Liquidity* + +So what is the problem? Seems legit and harmless enough?! Of course not - the biggest risk is Herd Mentality amplified Thousandfold by similarly trained Systems/Algorithms. If all the computers recognize the same signals (poor company earnings or poor economic outlook), and rush to Sell (or Buy) - the price impact can quickly escalate to Panic levels, causing even further snowball effect! + +While market system implemented *some* [Market circuit-breakers](https://www.investopedia.com/terms/c/circuitbreaker.asp) over the years, the risk is all still there. + +So why let machines trade stocks independently at all? Well, besides the fact it is realistically impossible to block, many argue it gives human traders certain liquidity. + +My thesis here is that this is a false promise - most of the automated systems have endless amount of rules and safeguards, and often in a situation of real "run-on-market" they step back and refuse to counter-trade also! In other words - if an average Joe decides they don't like "company X" any more and want to sell their shares ASAP, most likely a LOT of other people decided the same thing and automated systems detected this "avalanche" and would decide: let's see how low it goes before considering buying again! + +## The young "Flash Crash Trader" + +An interesting sidebar is a story of a brilliant young trader in England who noticed how several automated systems react to a particular Limit Order (order visible on the Markets L2 order book, but not yet executed). Many systems interpreted it as a certain Sentiment and were triggered to Buy/Sell ahead of it - while in fact the trader was "bluffing", and kept adjusting or cancelling his order shortly thereafter. [Here is the writeup from BBC](https://www.bbc.com/news/explainers-51265169) about him. + +## AI or KISS (Keep it Simple *Silly*)? + +So how does our enormous leap forward in AI affect these automated systems? Should we rush to regulate them before all our pensions are demolished by another "[Flash-Crash Trader Wiz Kid](https://www.bbc.com/news/explainers-51265169)"? + +My sense is that AI ultimately operates as our tool, as a sort of "faster assistant" for what human would have most likely done anyhow. In fact all the Quant funds have been revising their AI over the years, albeit sometimes after the trades, as a sort of HITL (Human-In-The-Loop) before the acronym gained popularity! A lot of market looks like self-fulfilling prophecies, especially around Technical Analysis patterns, which I also believe is a result of both human emotion actions and AI trained to recognize it amplify it via built-in expectations. + +This topic is huge, so I'll relent for now - below is some further reading, if you are interested. I also promise a post with a deeper dive into algorithmic trading in the future. + +Meanwhile, I recommend you Buy and Hold companies you personally like in your day-to-day - arguably the easiest and the best strategy over the long term! + +## Some further reading to explore + +- [What is Technical Analysis](https://www.investopedia.com/terms/t/technicalanalysis.asp) +- [Quant Funds that beat the stock market](https://medium.com/@siddharth.bhaisora/best-quant-funds-that-beat-the-stock-market-1597ca160df5) +- [A Random Walk Down Wall Street](https://www.amazon.com/Random-Walk-Down-Wall-Street-ebook/dp/B0B3G6FVT5/) + +![AI Robots trading on NYSE](multiple_AI_bots_NYSE_trading.png "AI Robots trading with each other on NYSE") + +{{< alert "image" >}} +**Images By DALL-E 3 from Microsoft Designer** +{{< /alert >}} +>> Prompt: Generate high resolution image of AI robots trading on New York stock exchange floor, sparks fly everywhere